ALG Dynamics · Stimulating Growth
Growth Marketing & Consumer Psychology · algdynamics.com
Stop betting. Start testing.
Here’s the pattern that quietly drains small-business marketing budgets: an idea feels exciting → money goes in → results are unclear → enthusiasm fades → new idea feels exciting. Repeat until the budget (or the owner) is exhausted.
The fix isn’t better ideas. It’s a better container for ideas: the growth experiment. Instead of asking “will this work?” (a bet), you ask “what will this teach me for the smallest possible cost?” (a test).
This reframe does something important psychologically, too. When an experiment “fails,” you didn’t fail — the hypothesis did, and you bought real information about your market at a discount. That mental distance is what lets you test boldly instead of defending your last idea long past its expiry date (we’re all wired to escalate commitment to things we’ve already sunk money into — the experiment format is your protection against your own wiring).
This planner gives you the full loop: choose → design → run → decide. One experiment at a time.
Step 1 — Choose what to test (the ICE score)
List your current marketing ideas. Score each 1–10 on three dimensions:
| Dimension | The question |
|—|—|
| Impact | If this works, how much does it move revenue or leads? |
| Confidence | How much real evidence do I have it’ll work — customer comments, competitor behavior, past results? (Gut feeling alone = 3 or below. Be honest.) |
| Ease | How cheap and fast can I test it? Doable inside 2–4 weeks with money I can afford to lose = high score. |
Average the three numbers. Highest score goes first. Run ONE experiment at a time — run three at once and you won’t know which one caused what.
Tip: beginners should over-weight Ease. Early on, the goal is learning the loop, not swinging for home runs.
Step 2 — Write the hypothesis (the sentence that keeps you honest)
Fill this in before you start — no exceptions:
We believe that [the change — e.g., offering a first-visit discount to neighborhood Facebook groups]
will result in [the outcome — e.g., new customer bookings]
because [your reasoning about the customer — e.g., locals want to try us but need a nudge to break their current habit].
We’ll know it worked if [specific number, by date — e.g., 10 bookings with the code within 3 weeks].
The two parts people skip are the two that matter most:
- The “because” clause — it forces you to state your assumption about why customers behave the way they do. When an experiment fails, it’s usually this assumption that was wrong, and knowing that is the real lesson.
- The success number, written in advance — after the fact, humans are gifted at reinterpreting any result as “pretty good, actually.” Deciding the bar beforehand is the only reliable defense.
Step 3 — Design the smallest honest test
- Shrink it. What’s the cheapest version that still produces a real answer? Don’t build the loyalty program — hand-stamp 30 cards. Don’t redo the website — change the headline on one page. Don’t produce a video series — post one video.
- Change ONE thing. New offer AND new audience AND new channel = a result you can’t interpret. One variable per experiment.
- Set the time box: 2–4 weeks. Shorter can’t accumulate enough signal; longer lets the experiment blur into “just something we do now.”
- Set the budget cap in advance — an amount you can lose without wincing. Write it down. No topping up mid-experiment “because it feels close.”
- Decide how you’ll count. A unique discount code, a dedicated link, a specific “mention this post” phrase, a tally sheet by the register. If you can’t count the result, redesign the test until you can.
- Expect small numbers — design around them. Most SMB experiments involve dozens of people, not thousands. That’s fine: you’re not running a clinical trial, you’re looking for a clear signal (roughly double-or-nothing differences), not a subtle one. If the result needs a statistician to detect, treat it as “no clear signal.”
Step 4 — Run it without sabotaging it
- Don’t peek daily. Early results mislead — the first five data points will swing your mood and tempt you to kill or crown the experiment prematurely. Check weekly, decide only at the end.
- Don’t tweak mid-flight. Changing the offer in week two means the results measure neither version. If you spot a genuine blunder (broken link, wrong price), fix it and restart the clock.
- Keep a 2-minute log. Once a week, jot: date, what happened, anything odd (weather, holiday, a competitor’s promo). Memory rewrites history; your log won’t.
- Collect the words, not just the numbers. If customers respond, note what they say — the exact phrases. What people say when they say yes is tomorrow’s ad copy.
Step 5 — Decide (the part everyone skips)
At the end date, compare the result to the success number you wrote in Step 2. Three outcomes, three moves:
| Outcome | What it means | Your move |
|—|—|—|
| Clear win (hit or beat the number) | The assumption in your “because” clause is probably right | Double down: repeat it bigger, make it permanent, or extend it to a neighboring audience/channel |
| Clear miss (nowhere near) | The assumption was wrong — that’s a finding, not a failure | Kill it cleanly and write one sentence: “We learned that [assumption] isn’t true because [evidence].” Then next idea |
| Murky middle (some signal, below the bar) | Often a right idea with wrong execution | One revision, one re-run — change the single element you most suspect (usually the offer or the headline). If it’s still murky after the re-run, kill it. Two strikes, out |
Then log it. Keep a running file — hypothesis, result, lesson, one line each. After six months you’ll own something priceless: a written record of what your specific market responds to. Most of your competitors are still guessing from scratch every quarter.
Twelve experiment ideas to seed your list
Score these with ICE against your own situation:
- A different headline on your most-visited page
- A first-time-customer offer with a trackable code
- Asking your 10 best customers for a referral, directly and personally
- A “mention this post” offer in one local Facebook group or subreddit
- Following up every quote/inquiry within 15 minutes for three weeks (speed test)
- A simple email to past customers who haven’t bought in 6+ months (“we miss you” + reason to return)
- Adding one honest trust element (real review, guarantee, your photo) to your main page
- A price-anchoring test: presenting your main offer next to a premium option
- A partnership pilot: one cross-promotion with a neighboring (non-competing) business
- Texting appointment/order updates instead of emailing them
- One educational post per week answering a question customers actually ask, for a month
- A “bring a friend” incentive for existing customers
The mindset that makes this work
You don’t need most experiments to win. You need a steady loop that keeps you learning faster than you spend — where wins get doubled and losses get capped early. Run one experiment a month and in a year you’ll have made 12 evidence-based decisions. That’s more than most businesses make in a decade of going with the gut.
Want a partner in the loop?
Designing sharp experiments — and reading the results without wishful thinking — is easier with someone who does it every week. We help SMB owners build a growth process grounded in how their customers actually think and decide, with the systems to run it without adding to your workload.
Let’s design your first experiment together: algdynamics.com/start-scaling
ALG Dynamics — Stimulating Growth
ALG Dynamics — Growth Marketing & Consumer Psychology. Want this working in your business? Start with a short conversation →